Genetic algorithms and non-invasive FECG enhancement

This thesis presents an approach to enhance a fetal electrocardiogram (FECG) via a non-invasive procedure using a genetic algorithm (GA). Many different techniques have been developed for FECG detection and enhancement, all of which have some limitations. The proposed method has been designed to overcome most of the difficulties of previous methods for FECG enhancement without distorting and/or adding misleading information to the FECG. The idea of the genetic algorithm is a completely new idea and has not appeared in prior art.

The genetic algorithm approach for FECG enhancement is based on subtracting a processed maternal electrocardiogram (MECG) from a processed abdominal ECG which contains an FECG and an MECG. The difference between straight subtraction and the genetic algorithm is the fact that the subtraction via a GA is a near optimal subtraction which eliminates the MECG in the abdominal signal.

The outcome from clinical data verified that the GA could enhance an FECG from an abdominal signal corrupted by an MECG and noise. Also, when the MECG and FECG are half-overlapped or perfectly overlapped, the GA still resolved the FECG. Finally, the most eloquent feature of the GA is the fact that an FECG can be resolved from an MECG with only two pairs of electrodes.